Functional Embryonic Cardiomyocytes after Disruption of the L-type α1C (Ca v 1.2) Calcium Channel Gene in the Mouse*

The L-type α1C(Cav1.2) calcium channel is the major calcium entry pathway in cardiac and smooth muscle. We inactivated theCa v 1.2 gene in two independent mouse lines that had indistinguishable phenotypes. Homozygous knockout embryos (Cav1.2−/−) died before day 14.5 postcoitum (p.c.). At day 12.5 p.c., the embryonic heart contracted with identical frequency in wild type (+/+), heterozygous (+/−), and homozygous (−/−) Cav1.2 embryos. Beating of isolated embryonic cardiomyocytes depended on extracellular calcium and was blocked by 1 μm nisoldipine. In (+/+), (+/−), and (−/−) cardiomyocytes, an L-type Ba2+ inward current (I Ba) was present that was stimulated by Bay K 8644 in all genotypes. At a holding potential of −80 mV, nisoldipine blocked I Ba of day 12.5 p.c. (+/+) and (+/−) cells with two IC50 values of ≈0.1 and ≈1 μm. Inhibition of I Ba of (−/−) cardiomyocytes was monophasic with an IC50 of ≈1 μm. The low affinity I Ba was also present in cardiomyocytes of homozygous α1D(Cav1.3) knockout embryos at day 12.5 p.c. These results indicate that, up to day 14 p.c., contraction of murine embryonic hearts requires an unidentified, low affinity L-type like calcium channel.

Calcium channels play an important role in the function of different tissues. The calcium entry via high voltage-activated (HVA) 1 calcium channels leads to excitation-contraction coupling in the heart, tension development in smooth muscle, neurotransmitter release in brain, and endocrine secretion in gland tissues. In the cardiovascular system, voltage-activated calcium channels are essential for the generation of normal cardiac rhythm, for induction of rhythm propagation through the atrioventricular node, and for the contraction of the atrial and ventricular muscle (1). In diseased myocardium, calcium channels can contribute to abnormal impulse generation and cardiac arrhythmias (2).
Calcium channels are hetero-oligomeric complexes of up to four subunits as follows: ␣ 1 , ␤, ␣ 2 ␦, and ␥ subunit. The ␣ 1 subunit contains the voltage sensor, the selectivity filter, the ion-conducting pore, and the binding sites for all known calcium channel blockers. The other subunits are auxiliary subunits, which modulate the channel function. Calcium channels can be further modulated by a variety of hormones, protein kinases, and protein phosphatases (3)(4)(5).
High voltage-activated calcium channels have been classified as L-type and non-L-type channels. L-type channels are encoded by four distinct genes, namely Ca v 1.1 to Ca v 1.4 (6), that give rise to numerous splice variants. Mammalian L-type channels have a similar ion selectivity and inactivation kinetics and are affected by dihydropyridines at similar concentrations. The expression pattern and the electrophysiology of L-type calcium channels have been studied extensively in pre-and postnatal heart cells of the mouse (7)(8)(9)(10). The major L-type channel expressed in the cardiac and smooth muscle is the Ca v 1.2 (11,12). In addition, the expression of the Ca v 1.3 gene was reported (13)(14)(15). However, the exact function of these channels often remained unclear. To analyze the functional relevance of the L-type Ca v 1.2 calcium channel for various tissues, two mouse lines were generated in which the Ca v 1.2 gene was disrupted at different exons. Both mouse lines had an identical phenotype. The homozygous (Ϫ/Ϫ) embryos died before day 14.5 p.c. Surprisingly, cardiomyocytes of 12.5-day-old p.c. embryos beat spontaneously using an unidentified L-type like calcium current.

EXPERIMENTAL PROCEDURES
Vector Construction (Mouse Line A)-Murine calcium channel Ca v 1.2 genomic DNA was obtained from a 129SV-P1 library (Genome Systems, St. Louis, MO) by screening with two primers amplifying 204 bp of the second exon of the Ca v 1.2 gene. Two exons were identified in the P1 clone coding for the second and third exon of mouse Ca v 1.2 gene (16,17). The third exon encoding part of the domain I of Ca v 1.2 was used for the construction of the targeting vector since alternative splicing has not been described for this exon. The key features of the targeting vector are shown in Fig. 1A. A neomycin resistance cassette (Neo) was placed into the MunI site of the third exon in the reverse direction of transcription. Additionally, a herpes simplex virus type I-thymidine kinase cassette (HSV-TK) was inserted 5Ј-terminal of the homologous region. Analysis of the genotype of the offsprings and proof for the correct insertion of the Neo were performed with primer pair NeoPA (5Ј-GCC TGC TCT TTA CTG AAG GCT CT-3Ј) and VS3 (5Ј-ACC ATT TGA AAT CAT TAT TTT ACT-3Ј) that amplify a 400-bp fragment of the mutated RNA and primer pair CSI (5Ј-ACG CCC AGC TCA TGC CAA CAT-3Ј) and mun3 (5Ј-TAA GGC CAC ACA ATT GGC AA-3Ј) that amplify a 354-bp fragment of the Ca v 1.2 exons 2 and 3 (Fig. 1C).
Vector Construction (Mouse Line B)-A second P1 plasmid was obtained from Genome Systems containing a different part of the murine Ca v 1.2 gene. Restriction map analysis and sequencing of this P1 plasmid revealed that exons 13-16 encoding part of repeat II of Ca v 1.2 were present (16,17). The key features of the targeting vector are shown in Fig. 2A. The vector contains a Neo and HSV-TK cassette and three 34-bp loxP sequences. Two of the loxP sites flank the Neo/HSV-TK cassette, which was inserted into the intron between exons 13 and 14. A third loxP site was placed into the intron between exons 15 and 16. To confirm the Cre-mediated deletion of the exons 14 and 15, RT-PCR was performed using the primer pair VS11 (5Ј-CTG GAA TTC CTT GAG CAA CCT TGT-3Ј) and VS16 (5Ј-AAT TTC CAC AGA TGA AGA GG-ATG-3Ј) to amplify exons 14 and 15 (329 bp) in (ϩ/ϩ), (ϩ/Ϫ), and (Ϫ/Ϫ) cells and the primer pair VS9 (5-ACA CAG CCA ATA AAG CCC TCC TG-3Ј) and VS18 (5Ј-GGC CAG CTT CTT CCT CTC CTT-3Ј) to amplify the sequence between exons 13 and 16 in the (Ϫ/Ϫ) mouse (341 bp).
Generation of Gene-targeted Mouse Lines-Sixty g of each targeting vector were linearized with NotI and electroporated into 1 ϫ 10 7 R1 embryonic stem (ES) cells obtained from Samuel Lunenfeld Research  A). A, a partial restriction map of the Ca v 1.2 (␣ 1C ) wild type locus (Aa), targeting vector (Ab), and targeted locus (Ac). Exons 2 and 3 are indicated as solid boxes and introns as solid lines. The Neo cassette was inserted into the MunI site of the third exon in opposite orientation. 5Ј to the Neo cassette, three stop codons were inserted in the three different reading frames. Double arrowhead lines in Aa and Ac represent the expected DNA fragments after SphI digest and hybridization with the EV500 probe (striped box in Ac). S, SphI; M, MunI; B, BamHI; A, AspI; C, ClaI. B, identification of Ca v 1.2 (ϩ/ϩ, wild type), (ϩ/Ϫ, heterozygous), and (Ϫ/Ϫ, homozygous) embryos by Southern blot analysis of SphI-digested DNA. Genomic DNA was derived from a single litter of 12.5 p.c. embryos from mating of heterozygous Ca v 1.2 (ϩ/Ϫ) mice. Hybridization with the EV500 probe yielded signals of 6.5 (ϩ/ϩ) and 4.5 kb (Ϫ/Ϫ). C, RT-PCR of RNA isolated from 12.5 p.c. embryos. PCR strategy to identify (Ca) wild type (WT) and (Cb) mutated (Ϫ/Ϫ) RNA. Cc, the primer pair CSI/mun3 amplified a 354-bp fragment from the wild type locus but did not amplify RNA from the knockout locus. Lanes 1 and 2, the RNA was reverse-transcribed; lane 3 (K), control plasmid containing the Ca v 1.2 cDNA; lanes 4 and 5, the RNA was not reverse-transcribed. The negative result of lane 5 shows that the amplicon of lane 2 was not derived from genomic DNA. Cd, the primer pair VS3/NeoPA yielded a 400-bp fragment only when the mutated locus is present. Lanes 1 and 2, the RNA was reverse-transcribed; lanes 3 and 4, the RNA was not reverse-transcribed. The negative result of lanes 3 and 4 shows that the amplicons of lanes 1 and 2 were not derived from genomic DNA. The location of the DNA fragment used as the 5Ј-hybridization probe in B is shown. Ab, the targeting vector contains a Neo/HSV-TK cassette and three loxP sequences (gray triangles numbered I-III). Two loxP sites flank the Neo/HSV-TK cassette, which is located in the intron upstream of exon 14. The third loxP site is located in the intron between exons 15 and 16. Ac, the targeted locus after homologous recombination. Ad, knockout locus after Cre-mediated excision of exons 14 and 15 and the Neo/HSV-TK cassette. The double arrowhead line in Aa and Ad shows the DNA fragment obtained after digestion with BamHI. A, Acc65I; B, BamHI; C, ClaI; EI, EcoRI; H, HindIII; SI, SstI. B, Southern blot analysis. DNA isolated from embryos at day 12.5 p.c. was digested with BamHI and then hybridized with the 5Ј probe yielding a 9-(ϩ/ϩ) and 17-kb (Ϫ/Ϫ) fragment. C, RT-PCR of RNA isolated from 12.5 p.c. embryos. Ca, primer pair VS11/VS16 amplifies a 329-bp fragment (arrow) of exons 14 and 15 in the wild type (ϩ/ϩ) and heterozygous (ϩ/Ϫ) but not in knockout (Ϫ/Ϫ) embryos. Cb, primer pair VS9/VS18 amplifies a 341-bp fragment (arrow) of exons 13-16 in RNA from homozygous knockout embryos (Ϫ/Ϫ). Left part, schematic drawing of spliced RNA. Right part, gels of PCR products; M, marker. D, the amplicon obtained by primer pair VS9/VS18 was sequenced and aligned with the murine cDNA sequence of Ca v 1.2 (16). Only part of the relevant sequence is shown. 1st line, exons and exon borders (͉). 2nd line (ϩ/ϩ), sequence of murine cDNA; only part of the sequence from exons 14 and 15 is shown (//, interruption of sequence). 3rd line (Ϫ/Ϫ), sequence of cDNA amplified from the RNA of a knockout embryo. Lowercase italic letters, sequence from the intron upstream of exon 16. *, stop codon caused by the frameshift.
Institute, Toronto, Canada. G418/ganciclovir-(mouse line A) and G418 (mouse line B)-resistant clones were screened by Southern blot analysis. Analysis of 311 G418-resistant clones revealed two clones (77 and 94 in mouse line B) that carried the floxed Neo/HSV-TK cassette and the third loxP site at the correct genomic region. 1 ϫ 10 7 ES cells of clone 77 were electroporated with 6 g of a Cre-expressing plasmid. Cells were plated at different dilutions and were selected with ganciclovir. Ganciclovir-resistant clones in which the Neo/HSV-TK cassette and exons 14 and 15 had been excised were identified by Southern analysis. ES cells carrying the disrupted allele (line A) or the Ca v 1.2 gene with the deleted exons 14 and 15 (line B) were microinjected into blastocysts from C57BL/6 mice to generate chimeric mice. Chimeric males were crossed with C57BL/6 females. Germ line transmission of the mutated Ca v 1.2 genes was verified by PCR analysis and Southern hybridization using tail DNA. The generation of the Ca v 1.3(Ϫ/Ϫ) mice has been described recently (18).
RNA Isolation and First Strand cDNA Synthesis-Total RNA was isolated from 12.5-day-old mouse embryos using TRIZOL LS Reagent (Life Technologies, Inc.). For the first strand synthesis, 4 g of total RNA were used according to the manufacturer's instructions. Primer pairs for the detection of the other calcium channels were as follows: Dissection and Cell Culturing of Murine Embryonic Cardiomyocytes-Individual embryos were obtained after breeding of heterozygous Ca v 1.2 (ϩ/Ϫ) mice at day 9.5 p.c. or later. Cardiac myocytes were isolated as described (8) at day 12.5 p.c. or later. Myocytes were plated on plastic coverslips and cultured in Dulbecco's modified Eagle's medium (8) supplemented with 10% fetal calf serum and 5% penicillin/ streptomycin (stock 10 mg/ml). Pharmacological tests were done in normal Tyrode's solution containing (in mM) 140 NaCl, 5.4 KCl, 1.8 CaCl 2 , 1 MgCl 2 , 10 glucose, 5 HEPES, pH 7.4. Genotyping of embryos was done by PCR.
Electrophysiology-Whole-cell currents were recorded at room temperature 18 -48 h after plating using fire-polished electrodes with resistances of 2-3.5 M⍀. Pipettes were filled with (in mM) 60 CsCl, 50 aspartic acid, 68 CsOH, 1 MgCl 2 , 5 K-ATP, 1 CaCl 2 , 10 HEPES, 11 EGTA, pH 7.4. Extracellular solution for sealing and recording of sodium current (I Na ) was (in mM) 130 NaCl, 4.8 KCl, 5 BaCl 2 , 1 MgCl 2 , 5 glucose, 5 HEPES, pH 7.4. To isolate barium currents (I Ba ), the bath solution was changed to (in mM) 130 N-methyl-D-glucamine, 4.8 CsCl, 5 BaCl 2 , 5 glucose, 5 HEPES, pH 7.4. The holding potential (HP) was Ϫ80 mV. Trains of test pulses were to Ϫ40 mV for I Na or to 0 mV for I Ba of L-type calcium channel applied once every 10 s for 40 ms. Data were collected and stored at an EPC-9 computer under control of Pulse software (HEKA electronics). Total cell membrane capacitance was determined by compensation mechanisms of the EPC9 computer and used as a measurement of membrane area. (ϩ/ϩ), (ϩ/Ϫ), and (Ϫ/Ϫ) cardiomyocytes had similar capacities of 30 Ϯ 2.6 (n ϭ 60), 26 Ϯ 2.3 (n ϭ 58), and 26 Ϯ 2 (n ϭ 53) pF, respectively. Inactivation curves were fitted by a Boltzmann relation as follows: where I is the current, I max is the maximal current at the beginning of the experiment, V is the potential, V 0.5 is the midpoint of the curve, k is the slope factor, and A is the non-inactivating part. Ba 2ϩ /Ca 2ϩ selectivity of the current was determined by a 100-ms pulse from Ϫ80 mV (HP) to 0 mV for (ϩ/ϩ) and (ϩ/Ϫ) and to Ϫ10 mV for (Ϫ/Ϫ) cardiomyocytes at 0.2 Hz. Five mM Ba 2ϩ was exchanged for 5 mM Ca 2ϩ in the bath solution. In some experiments the sequence was reversed.
Cumulative dose-response curves were recorded using 2-3 different nisoldipine concentrations per cell. The number of experiments was 4 -9 for each concentration. The stoichiometries and apparent affinities of nisoldipine were determined by fitting the averaged dose-inhibition points to the Hill equation: is the concentration of nisoldipine, IC 50 is the half-blocking concentration, and H is the Hill coefficient, I is the averaged current measured at any concentration of nisoldipine, and I max is the average current measured in the absence of nisoldipine. To obtain apparent affinities for complex dose-inhibition relations, sums of Hill terms similar in form to that described above were fitted to the data.
Stock solutions were as follows: Bay K 8644 10 mM in ethanol; nisoldipine 20 mM in ethanol; isoproterenol ϩ ascorbic acid 10 mM each in H 2 O; tetrodotoxin citrate (TTX) 1 mM in H 2 O. When required, stock solutions were freshly diluted to the indicated concentrations with the used extracellular solution. Data are shown as mean Ϯ S.E. with the number of cells in parentheses. Graphics and statistical data analysis using Student's t test were carried out using ORIGIN software (Microcal).

RESULTS
Genotype-Two different mouse lines were generated in which the Ca v 1.2 calcium channel gene was disrupted. In mouse line A, a neomycin resistance cassette was inserted into the third exon of the Ca v 1.2 gene (Fig. 1A) in the opposite orientation. 5Ј to the Neo cassette stop codons were inserted into each reading frame of Ca v 1.2. The generation of functional channels is highly unlikely because the introduced modification leads to a truncated non-functional protein also in the case of an incorrect splicing, i.e. even if exon three is skipped, because of an early stop codon in exon four. Southern hybridization (Fig. 1B) (Fig.  2Cb). The primary transcript was spliced from exon 13 to an intron sequence directly upstream of exon 16 (Fig. 2D). The new splicing event caused a frameshift resulting in an early stop codon. In agreement with these results, Western analysis with antibodies against the Ca v 1.2 and the Ca v 1.1 protein yielded no specific bands in (Ϫ/Ϫ) embryonic hearts.
Identical results were obtained with both knockout lines. All experiments were at least repeated once in the other knockout line. Heterozygous Ca v 1.2 (ϩ/Ϫ) mice were indistinguishable from wild type (ϩ/ϩ) mice in shape, development, and behavior. The mating of heterozygous mice led to viable (ϩ/ϩ) and Phenotype-Visual inspection suggested that Ca v 1.2 (ϩ/ϩ), (ϩ/Ϫ), and (Ϫ/Ϫ) embryos developed normally up to day 12.5 p.c. Hearts contracted with the same frequency at day 12.5 p.c. (Fig. 3A). After day 14.5 p.c., the beating frequency of the remaining (ϩ/ϩ) and (ϩ/Ϫ) embryos increased. Cardiac cells from day 12.5 p.c. (ϩ/ϩ), (ϩ/Ϫ), and (Ϫ/Ϫ) embryos could be cultured for more than a week. During this time, the frequency of spontaneous contractions increased in each genotype from about 30 to around 160 beats/min (Fig. 3B) suggesting that the Ca v 1.2 gene is not necessary for rhythmic activity or is compensated during embryonic development but is required after day 13 p.c. A previous report (9) indicated that the spontaneous contractions of cardiomyocytes from stage II embryoid bodies were caused by oscillations of intracellular [Ca 2ϩ ] and did not require the influx of extracellular Ca 2ϩ during each beat. To confirm these results, (ϩ/Ϫ) and (Ϫ/Ϫ) heart cells were superfused with Ca 2ϩ -free normal Tyrode's solution. Within 6 s, all cardiomyocytes stopped contracting. Addition of 1.8 mM Ca 2ϩ restored beating in 2-5 s (n ϭ 7 experiments for each genotype isolated from 3 different embryos). Repeated cycles of Ca 2ϩ withdrawal and re-addition stopped and started contractions each time, respectively. Superfusion of (ϩ/Ϫ) or (Ϫ/Ϫ) cardiomyocytes with the dihydropyridine (DHP) nisoldipine (1 M) stopped beating in 0.5 min (n ϭ 8). The inhibitory effect of nisoldipine was reversed by washout of the compound. These results suggested that the rhythmic activity of day 12.5 p.c. heart cells was not caused by intracellular [Ca 2ϩ ] oscillations but depended on the influx of extracellular Ca 2ϩ through a channel with the characteristics of an L-type calcium channel. The nature of this putative channel was analyzed by the patch clamp technique using isolated cardiomyocytes.
Maximal inward currents with Ca 2ϩ as charge carrier were reduced in each genotype, and current inactivation was increased to 90% in the presence of Ca 2ϩ (Fig. 4, D and E) suggesting Ca 2ϩ -dependent inactivation of the channel in each genotype.
Superfusion of individual cells with the calcium channel agonist Bay K 8644 (1 M) increased I Ba and induced a shift of the I-V relation to hyperpolarized potentials in all three genotypes. The calcium channel blocker nisoldipine (1 M) inhibited I Ba in each cell line but with less efficiency in (Ϫ/Ϫ) cardiomyocytes (Fig. 4). The same results were obtained in mouse line A and B (Fig. 4, A and B). However, the current amplitude differed significantly between the three genotypes (Fig. 4C). I Ba increased slightly in the (ϩ/Ϫ) cells after day 14.5 p.c. and was equal to I Ba of (ϩ/ϩ) cells. The difference in current densities was not due to distinct cell sizes or differences in I Na . The I Na amplitudes were similar with 251 Ϯ 19 (n ϭ 57) for (ϩ/ϩ), 219 Ϯ 18 (n ϭ 54) for (ϩ/Ϫ), and 208 Ϯ 16 (n ϭ 61) pA/pF for (Ϫ/Ϫ) cells. This analysis suggested that embryonic cardiac cells from two independent Ca v 1.2 knockout mouse lines expressed a bona fide L-type calcium channel. An alternative explanation for this phenotype was that the observed L-type channel was the so-called slip-mode sodium conductance channel (19). This channel is blocked by TTX with an IC 50 of 0.1 M and allows permeation of calcium in the presence of cAMP kinase or after activation of the ␤-adrenergic receptor. In support, I Ba was stimulated in each cell line 1.8 -2.0-fold (n ϭ 8 to 20 cells) by isoproterenol. A similar adrenergic stimulation of I Ba has been reported for day 9.5 p.c. mouse embryonic heart cells (10). However, I Ba was not affected at all by 10 M TTX, whereas I Na was blocked reversibly in each cell line (not shown). Therefore, it was concluded that the slip-mode channel did not cause the observed DHP-sensitive I Ba in Ca v 1.2(Ϫ/Ϫ) cardiac cells.
DHP Sensitivity of I Ba -The experiments shown in Fig. 4 indicated that I Ba of the Ca v 1.2(Ϫ/Ϫ) cells was less sensitive to nisoldipine than that of the cells with a wild type or heterozygous genotype. Therefore, the extent of channel block was tested by superfusion of the (ϩ/ϩ), (ϩ/Ϫ), or (Ϫ/Ϫ) cells with 1 M nisoldipine at the HP of Ϫ80 mV with trains of test pulses (Fig. 5A). Nisoldipine reduced I Ba of (ϩ/ϩ) and (ϩ/Ϫ) cardiomyocytes to 23 Ϯ 3.7% (n ϭ 10) and 27 Ϯ 3.3% (n ϭ 8), respectively. In contrast, I Ba of (Ϫ/Ϫ) cells was reduced only to 65 Ϯ 3.3% (n ϭ 11) of the control. A shift of the HP from Ϫ80 to Ϫ40 mV reduced I Ba to zero in all three genotypes. I Ba recovered to 80% of the previous value in each cell line after reversal of the HP from Ϫ40 to Ϫ80 mV. The voltage-dependent reversibility of the block indicated that nisoldipine bound preferentially to the inactivated state of the channel, a phenomenon described extensively for the Ca v 1.2 channel (see Refs. 3, 5, and 20 and references cited therein). The affinity of nisoldipine to block I Ba at a HP of Ϫ80 mV was determined from dose-inhibition curves (Fig. 5B) (Table I). The inhibition curves were well fitted with a Hill coefficient of 1.0 suggesting that nisoldipine blocked two independent currents in the (ϩ/ϩ) and (ϩ/Ϫ) cells. Considering the relative variability introduced by the calculation method, we suggest that the high IC 50 values were not different between the three genotypes and that the real high IC 50 value is close to 1 M.
L-type Ca v 1.2 channels have been reported to be blocked by nisoldipine with IC 50 values around 10 nM (21) suggesting that day 12.5 p.c. wild type myocytes expressed a relative low affinity L-type Ca v 1.2 channel. This low affinity is apparently a developmental property of the mouse heart, since day 15.5 p.c. wild type myocytes were blocked by nisoldipine with IC 50 values of 10 nM and 5.7 M (Fig. 5B and Table I). The high (10 nM) and intermediate (100 nM) DHP sensitivity was most likely caused by the expression of different splice variants of the Ca v 1.2 gene (21)(22)(23)(24)(25). Ca v 1.2 channels that contain the sequence of exon 21 have a lower affinity for DHPs than those containing the alternatively used exon 22 (23,25). In agreement with the increase in the affinity between day 12.5 and 15.5 p.c., the relative abundance of exon 22 mRNA increased and that of exon 21 decreased in wild type cardiomyocytes  between these days. These results confirmed that embryonic cardiomyocytes express L-type Ca v 1.2 channels with different DHP sensitivity in the nanomolar range. In addition to the Ca v 1.2 channel, embryonic cardiomyocytes express a second L-type like calcium channel which is also present in the Ca v 1.2 (Ϫ/Ϫ) cells and has a nisoldipine affinity in the molar range. The current of (ϩ/ϩ) and ( In addition, Ca V 1.4 mRNA was detected in the adult eye but not the heart of embryos by in situ hybridization with a Ca V 1.4specific probe (data not shown). The expression of Ca v 1.1 has been reported previously in embryonic heart cell lines (28,29). However, the possibility that Ca v 1.1 caused the rhythmic activity was rejected, since no specific protein band was detected by Western blot and the fast activation kinetics of the low affinity L-type like channel were not in line with those of a skeletal muscle calcium channel (see Fig. 4). Adult hearts express the Ca v 1.3 channel (13)(14)(15). Deletion of the Ca v 1.3 gene caused cardiac arrhythmia (18). Therefore, day 12.5 p.c. Ca v 1.3 (Ϫ/Ϫ) heart cells were analyzed. These cells had a regular Ca v 1.2 L-type I Ba that was blocked half-maximally at 3 nM nisoldipine ( Fig. 5B and Table I). However, these cells had still the second I Ba that was blocked half-maximally at 0.7 M nisoldipine, which value is very close to the IC 50 value of Ca v 1.2 (Ϫ/Ϫ) cells. These findings strengthen the hypothesis that the low affinity L-type like I Ba was caused by a channel not identified so far. DISCUSSION Five conclusions can been drawn from this study as follows. The L-type like channel cannot be an alternatively spliced product of the Ca v 1.2 gene. In mouse line B, exons 14 and 15 were deleted that code for part of the channel pore. Thus, a hypothetical channel would not contain a pore and should, therefore, be non-conducting. Furthermore, the aberrant RNA splicing caused a frameshift that would yield a truncated channel with no pore region. In mouse line A, an unexpected splicing event from exon 2 to exon 4 would lead to a frameshift and stop in exon 4. The L-type like channel is also not coded for by the Ca v 1.3 gene, since it was still present in Ca v 1.3 knockout embryos. The mRNA of the skeletal muscle Ca v 1.1 channel was detected in embryonic mouse cardiac cells. Several lines of evidence suggest that the L-type like channel was not the Ca v 1.1 channel. 1) The activation kinetics of this channel were much faster than those reported for the calcium channel of developing murine skeletal muscle (30). 2) Mice in which the Ca v 1.1 gene is disrupted die at birth but develop normally until birth (30). Therefore, it is very unlikely that the L-type like current was caused by the Ca v 1.1 channel. This current was also not caused by the Ca v 1.4 channel. This channel has been reported to be specifically expressed in the retina (33,34). In line with this expression pattern, the mRNA of the Ca v 1.4 gene was not detected in the embryonic heart. Its sensitivity against nisoldipine distinguishes the L-type like channel from the brain channels Ca v 2.1, Ca v 2.2, and Ca v 2.3. Furthermore, disruption of the Ca v 2.1 and Ca v 2.3 gene leads to viable pups (31,32) arguing against an essential role of these channels for cardiac rhythm generation.
The low affinity for nisoldipine would be in line with reports that some T-type calcium currents are blocked at high concentrations by several DHPs (35). Ca v 3.1 and Ca v 3.2 are expressed in the heart (36). However, the Ca 2ϩ -dependent and slow inactivation of the L-type like channel has not been observed with T-type channels. Furthermore, the expressed Ca v 3.1 channel is inhibited at submicromolar concentrations of mibefradil (37), is not activated by Bay K 8644, and is inhibited marginally by 1 M (ϩ)-isradipine or 10 M nifedipine (37). Similarly, amlodipine blocked the expressed Ca v 3.2 channel with an IC 50 of 31 M (38). These considerations strengthen the notion that the Ltype like current was caused by an unknown calcium channel.
The preliminary characterization of the new current showed that it has many properties of a classical HVA L-type calcium channel. The current differs from the classical L-type calcium channel by its low affinity for the DHP nisoldipine at negative membrane potentials but resembles Ca v 1.2 channels by the voltage dependence of the block (see Fig. 5A). This property was responsible for the inhibition of cardiac contraction by 1 M nisoldipine, since at depolarized membrane potentials which are necessary for channel opening the affinity for nisoldipine increased significantly. The affinity of the Ca v 1.2 channel for DHPs is lowered 100-fold by mutation of Tyr-1485, Met-1486, and Ile-1493 in the IVS6 segment (39) and is lost upon mutation of Thr-1061 in the IIIS5 segment (see Refs. 3,5,and 20). It is possible that the new channel has an altered IVS6 segment but retained other parts of the DHP-binding site.
In conclusion, the present study shows that early cardiac rhythm generation required an unidentified L-type like calcium channel. This channel has many properties of the well characterized Ca v 1.2 channel. Presumably, this similarity has prevented so far its identification in embryonic cells. The channel was also present in fetal hearts and may be present in adult hearts. Its functional significance beyond embryonic development remains to be established and will require the identification of its structure. The Drosophila melanogaster and Caenorhabditis elegans genome contain calcium channel genes of unknown function (40). One may speculate that the L-type like channel is encoded by a similar gene in the mouse.